Running locally for Church Translation

#22
by shapmeister - opened

Hello hugging face community, I have tried to take a dive installing seamless streaming locally on a fairly robust Ubuntu machine with an RTX 3090 and 24gb RAM. I have made it close several times, and landed at multiple errors when trying to start the server that are due to component version mismatches. Following instructions in the readme, when I install either fairseq2 or the requirements.txt portion, Pytorch, torch audio, and others change their installed versions. My machine is capable of up to cuda 12.4. Does someone have an updated readme and instructions to make this install go seamlessly? I have probably spent a good 50 hours on this thing, but I am determined to get it right. Any help would be very much appreciated.

Sign up or log in to comment